M PRA Munich Personal RePEc Archive Empirical Evidence on the Long-Run Money Demand Function in the GCC Countries Helmi Hamdi and Rashid Sbia and ali said 2014 Online at http://mpra.ub.uni-muenchen.de/63306/ MPRA Paper No. 63306, posted 29. March 2015 17:51 UTC Empirical Evidence on the Long-Run Money Demand Function in the GCC Countries Helmi Hamdi Aix-Marseille University CERGAM (4225), France. Email: [email protected] Rashid Sbia Corresponding author DULBEA, Solvay Brussels School of Economics and Management, Belgium. Email. [email protected] Ali Said Strayer University, United States. Email: [email protected] ABSTRACT: The broad aim of this paper is to estimate the money demand function for the case of six Gulf Cooperation Council countries. By applying panel cointegration tests, the empirical results reveal strong evidence of cointegration between the variables of the model for individual countries as well as for the panel. Moreover, the results support the existence of a stable money function in the long-run estimation. The Granger non-causality test due to Toda and Yamamoto (1995) procedure shows evidence of a bidirectional causal relationship between money demand and income for panel estimation. At an individual level, the results change from one country to another one. Keywords: Money Demand; GCC, Panel Cointegration; Toda-Yamamoto JEL Classifications: C22, C23, E52, E41, F41 1. INTRODUCTION Since the pioneering work by Friedman (1956), the money demand function has received a great deal of attention by scholars, policymakers and governors. This is mainly due to the role of money in economy, notably in the implementation of monetary policy. Since a long time, Central bankers have used money demand to control inflation through the appropriate and adjustment of the money supply (Hayo, 1999). Money demand function provides information about the portfolio distribution (Duca and VanHoose 2004) and it plays an important role in creating an efficient and effective monetary policy strategy (Friedman, 1959; Friedman and Schwartz, 1982; Laidler, 1977; Laidler, 1982). The monetary policies can be suitable to give a clear indication about inflation in medium and long term. Kaldor (1982) showed that money supply is considered of being a causal in the process of inflation. However, Valadkhani (2006) showed the importance of not focusing on a single policy instrument and neglect other information and variables such interest rate. Hence, interest rate and monetary aggregates are both important to select an effective monetary policy action. Further studies that have examined the wealth variable as other variable that influence money demand such as (Boone and van den Noord 2008; de Bondt 2009; Dreger and Wolters 2010). Kumar, Webber and Fargher (2010) analyzed the level and stability of money demand in Nigeria between 1960 and 2008. The study showed that Nigeria could efficiently use the demand of money as an instrument of monetary policy. Recently, there has been a great resurgence of interest in the issue of the role of money demand function in conducting effective monetary policy. One symptom of this phenomenon is the huge academic research on the topic investigating the demand of money at both single country level and panel or groups of countries. Hence, the aim of this study is to investigate the importance of money demand in conducting a sound monetary policy in the Gulf Cooperation Council Countries. We are interested in GCC area for different reasons. First, during the past decade GCC countries have been witnessing an unprecedented economic performance thanks to the windfall of oil revenues. The growth was on par with other emerging markets with an average rates exceeding 5-6% and much faster than advanced economies. Second, GCC governments have adopted development strategies that prioritize the modernization of their financial systems within a large economic diversification plan (Hamdi et al 2014). Third, the region as a whole has become a hub of finance, notably center of Islamic finance and Islamic insurance, and the preferred destination of international financial companies (Hamdi and Sbia 2013). Fourth, these countries have fixed their national currencies in the past to the US dollar and they are planning to move toward a single currency in the few coming months. The GCC regional bank was already implemented in Riyadh and it started executing the required first step for the formation of monetary Union. In addition, those countries have started a custom union and a common market grants national treatment to all GCC firms and citizens in any other GCC country, which by doing so, they have removed all barriers to cross country investment and services trade which in an effort moving closer and closer towards an economic union for those counties. The remainder of the paper is organized as follows. Section 2 provides a brief review of literature, section 3 gives data description and methodology, section 4 presents the results and finally section 5 concludes. 2. LITERATURE REVIEW Literature on money demand function is rich and huge. Researchers investigated this topic at both country level case study and panel of countries. For example, at single country level, Ghartey (1998) investigated the demand for money in Ghana using the Engle-Granger (1987) and Johansen’s (1988) cointegration and error-correction modeling approach. The results showed that money demand in Ghana is stable. Cheong (2003) examined the impact of financial liberalization on demand of money stability in Korea. Unlike the previous studies, he demonstrated that the stability of money demand function has not been affected by the set of financial measures over the sample period. However, the author suggested that results should be taken carefully as they depend heavily on statistical properties of the model and the suitable interpretation. He concluded that the ECM model is interpreted solely as backward-looking and is also invariant to shifts in regime and various policy reforms. Lee and Chien (2008) showed that money demand in China has a significant effect on the economic and financial stability, while Baharumshah et al. (2009) and Wu (2009) demonstrate that a stability of money demand function would exist as long as there is proper accounting in use. Recently, Jawadi and Sousa (2013) estimated money demand equations for the euro area, the US and the UK using a quantile regression framework and a smooth-transition regression. The two approaches provided different findings. The quantile regression approach revealed that the income and the interest rate semi-elasticities are meaningfully different from the OLS estimates and the reaction of money demand to inflation tends to be greater when real money holdings are particularly low. The smooth transition model revealed also two motivating results. First, it captured soundly the nonlinear dynamics of the money demand function. Second, it showed that the elasticity of money demand with respect to inflation rate, interest rate, GDP and exchange rate diverges not only according to the regime considered, but also across the countries chosen in the sample. In comparison to a large number of empirical works on the demand for money for other countries and group of countries, there are only a handful of empirical studies on GCC countries. At single country level, Darrat and Mutawa (1996) have used the cointegration and error-correction model to measure the money demand in the United Arab Emirates. The result of the study confirmed the support of the use of M1 as an intermediate target for monetary policy in the United Arab Emirates. Khatib and Towaijari (1999) have studies money demand in Saudi Arabia. They regressed the log of real M1 on the log of non-oil GDP, local interest rate, expected inflation rate and real exchange rate during the instable period of 1977-1997. They used the residuals to estimate an error correction model. The result showed that influence of the interest rate is low and statistically insignificant and the researchers explained that due to Islamic values and cultural in Saudi Arabia. In panel framework, Harb (2004) found that cointegration between money and non-oil GDP for the period of 1979-2000. The study has used Pedroni’s (1999) panel cointegration method. The study found significant negative the semi elasticity of money demand in connection to interest rate. The other study by Lee et al. (2008) estimated money demand function for six selected countries of the Gulf Cooperation Council (GCC) for the same period of Harb (2004) using Likelihood-based cointegration tests in heterogeneous panels. The study findings were at least two cointegrated correlation in the four- 2 dimensional vector error-correction model for the variables of the real money balance, the real scale variable, the nominal interest rate, and the exchange rate. Basher and Fachin (2012) estimated the long-run demand for broad money at the Gulf Cooperation Council area level and at single country level over the 1980–2009 period using times series and panel techniques. First results confirmed the stability of money demand in the long-run both nationally and regionally. Further, the estimated income and interest elasticities in Qatar, Saudi Arabia and the UAE offered an authentication for the Baumol-Tobin version of the inventory analysis of the transactions demand for money. However, income elasticities in the other GCC economies reﬂected portfolio demand more strongly than transaction demand with lower interest rate (semi-)elasticities. They discussed how the movements in income velocity could resolve the varying elasticities documented across the six countries. 3. DATA AND METHODOLOGY 3.1. Data We follow the pioneering works on money demand function (Arango and Nadiri (1981) Laidler (1985), Hoffman and Rasche (1991), Miller (1991), Baba et al. (1992), Stock and Watson (1993), Mehra (1993), Ball (2001), Mark and Sul 2003, Dickey et al. (1991), Miller (1991), Mehra (1992), BahmaniOskooee and Shabsigh (1996), Valadkhani and Alauddin (2003), Harb (2004), among others) in which a basic representation of the long-run money demand can be described as follows: Md f YP , rs P ( ) ( ) (1) Md Where represents real money proxied by M2, where nominal money stocks M d are deflated by P the CPIs ( Pi ,t ) ; (Yi ,t ) is the scale variable proxied by the country’s income, (ri ,t ) is a domestic interest rate1 which represents the opportunity cost of holding money. Our sample covers the six GCC countries i;e Bahrain, Kuwait, Oman, Qatar, Arabia and UAE and data used is quarterly and covers the period 1980Q1-2011Q4. Data was obtained from different sources such as the International Financial Statistics (IFS), the World Bank (2012) as well as the Arab Monetary Fund statistical book. We use non-oil GDP (Y) as a scale variable for the 4 most oil producing countries: Kuwait, Qatar, Arabia and UAE and GDP for Oman and Bahrain. Empirical papers mainly rely on equation (1), but in many cases researchers employ an augmented money demand function by adding some variables (Foresti and Napolitano 2012). In our case, as GCC countries are open economies, therefore we will add to equation 1 a foreign opportunity cost of holding domestic money in the GCC countries proxied by two indicators which are the UK threemonth Treasury-Bill rate and the US robiL rate. Moreover, unlike Darrat and Al-Sowaidy (2009) our money demand equation includes the exchange rate variable. Exchange rate is the amount of the local currency per one unit of SDR (Harb 2003). In fact, as the GCC currencies are highly linked to the US economy through the fixed exchange rate (peg), therefore, any depreciation or depreciation of the US dollar would impact automatically the local currencies of GCC countries. The inclusion of exchange rate variable in the standard function of money demand is first suggested by Mundell (1963) and later by the works of Bahmani-Oskooee (1996), Bahmani-Oskooee and Techaratanachai (2001), Harb (2004), Bahmani-Oskooee and Tanku (2006). According to what cited above, the money demand function could be expressed as follows Md f YP , rs , E s Tbill s Libors P ( ) ( ) ( ) ( ) ( ) (2) 1 Given the lack of data in GCC countries, we followed Harb (2004) and we proxied interest rates by average time deposit rates. 3 Where E is the exchange rate variable. Tbill and Libor are the UK three-month Treasury-Bill rate and the US robiL rate. The volatility of E leads to volatility of the domestic currency against foreign currency (or SDR). The estimation of the semi-logarithmic linear specification of long-run money demand takes the following form. In empirical analyses, is typically preferred. ln Md 0 1 ln yi ,t 2 rs ,i t P (3) i refers to specific effects in a country, 1 is the income elasticity, and 2 is the interest rate semi-elasticity; for i = 1, 2, . . . N; t = 1, 2, . . . T; where N = 6 and T = 124 which gives us 6* 31= 744 observations. ε represents the error term. The augmented money demand function is expressed as follows Md ln 0 1 ln yi ,t 2 rs ,i 3 ln Et ,i 4Tbill i ,t 5 Libori ,t t P (4) Economic theory reveals that scale variable should have a positive effect on money holdings. Therefore, the income elasticity coefficient i is expected to be positive. Regarding the opportunity cost, it should have a negative impact on money demand; thus 2 is expected to be negative. For the elasticity coefficient on the exchange rate variable 3 it can be either positive or negative (Arango and Nadiri, 1981). In fact, a depreciation of exchange rate is associated with an increase in income which in turn would rise of domestic money. In this case, the coefficient of exchange rate is positive. However, an appreciation in exchange rate is associated with a decrease in domestic money demand (currency substitution. In this case, we could expect a negative sign of the coefficient of exchange rate. 3.2. Methodology Our empirical study is divided in three steps. The first step is to test whether the variables contain a panel unit root to confirm the stationarity of M2, NOG (or GDP), Drate, Tbill, and Xrate. This is done by performing five type of panel unit root tests which are: Levin-Lin-Chu (LLC, 2002), Im, Pesaran and Shin (IPS, 2003), the Augmented Dickey–Fuller (F-ADF), Philips–Perron (PP, 1998) and finally Breitung (2000). The second step is to check for panel cointegration tests using Kao (1999) and Pedroni (2004) to establish a cointegrating long-term equilibrium relationship between money demand and its determinants. Finally, the third step, we test for panel cointegration by using three different techniques: fully-modified ordinary least squares (FM-OLS) of Phillips and Hansen (1990) Kao and Chiang (2000), and Pedroni (2004) dynamic ordinary least squares (DOLS) estimator of Stock and Watson (1993) and canonical cointegrating regression (CCR) proposed by Park (1992). 4. EMPIRICAL RESULTS 4.1. Panel unit roots and panel cointegration tests The properties of the five variables’ time series are verified through the use of four types of panel unit root tests for balanced GCC panel data. These tests are the LLC, Breitung, IPS and F-ADF panel unit root tests. The two former tests assume that there is a common unit root process across crosssections while the alternative hypothesis does not have a unit root. However, the two later tests assume that there are individual unit root processes across the cross-sections, while the alternative hypothesis of some cross sections does not contain a unit root. The results of the LLC, Breitung, IPS and F-ADF panel unit root tests for each of the variable are displayed in table 1. We conducted each test for the level and first difference of each variable. The results show that the series are likely to contain a panel unit root in their levels. However, when applying each variable at first difference of the panel unit root test, all tests reject the null hypothesis at the 1% level of significance indicating that they are integrated at order one ,i.e., I(1). Table 1. Panel Unit Root Test Variable Lnm2 LnY LLC Level 2.195 -1.993 1st diff -7.201*** -4.433*** BREITUNG Level 0.092 -2.481*** 1st diff -1.493* -2.035** IPS Level -1.889** -2.683*** 1st diff -8.012*** -7.558*** F-ADF Level 50.205*** 37.725*** 1st diff 150.24*** 77.868** F-PP Level 182.57*** 35.928*** Order of int. 1st diff 194.567*** 123.659*** 4 I(1) I(1) Drate Tbill Ln Xrate -0.171 -1.679 -0.715 -4.925*** -4.803*** -8.001*** -2.622*** -2.241** -3.990*** -4.527*** -2.892*** -1.878** -1.880** -5.477*** -1.314* -4.260*** -7.334*** -6.302*** 18.973** 47.194*** 12.965 37.920*** 68.674*** 56.387*** 14.204 17.288* 8.6788 145.512*** 263.397*** 292.061*** Notes: All tests examine the null hypothesis of non-stationary. The tests are: Levin, Lin and Chu, 2002 (LLC); Breitung, 2000; Im, Pesaran and Shin, 2003 (IPS); ADF Fisher (ADF); PP Fisher (PP) due to Maddala and Wu, 1999. *** indicates statistical significance at the 1% level. The optimal lag length is selected automatically using the Schwarz information criteria (SIC). Probabilities for the ADF (Fisher Chi-square) and PP (Fisher chi-square) tests are computed using an asymptotic χ2 distribution. All other tests assume asymptotic normality. Following Bacher and Faccini (2012), as Libor is identical for all GCC countries; therefore we can test the statistionarity of the variable using the Augmented Dickey–Fuller (ADF) and Philips and Peron (PP) unit root tests. The results are displayed in table below and they show that we cannot reject the null hypothesis of unit roots for Libor in level forms. However, the null hypothesis is rejected when the ADF and PP tests are applied to the first differences indicating that libor is integrated of order one, I(1). Libor Level -1.5775 Table 2. Unit Root Test ADF First Level -10.1552*** -2.2236 PP First -10.1646*** *** indicates statistical significance at the 1% level. After checking the integration of our six variables at order one, I(1), the Pedroni, Kao and Fisher tests for balanced (GCC) panel date are used in order to verify the presence of a long-run relation between the variables in our dataset. The test results of Pedroni displayed in table 3. Table 3. Results of the Balanced Panel Cointegration Tests for GCC Countries Statistic 2.926981 Panel v-Statistic Weighted Statistic Panel rho-Statistic Weighted Statistic Panel PP-Statistic Weighted Statistic Panel ADF-Statistic Weighted Statistic -3.529681 -3.186576 -2.227194 Statistic Group rho-Statistic Group PP-Statistic Group ADF-Statistic Kao Test. ADF Johansen Fisher Panel Cointegration Test Null Hypo. Max-Eigen. r=o 70.74 (0.000)*** r<1 12.02 (0.284) r<2 8.197 (0.609) r<3 7.561(0.406) -2.664054 -2.943584 -2.587459 Prob. 0.0017 0.0002 0.0007 0.9731 Prob. 0.0039 0.0016 0.0048 1.978329* (0.0239) Trace 70.16 (0000)*** 18.95 (0.240) 12.05(0.281) 9.475(0.487) Note: The optimal lag lengths are selected using SBC. Figures in parenthesis are probability values. Trace test and Max-eigenvalue test indicate 1cointegrating vector at the 0.01 level *** Denotes the rejection of the null hypothesis at 1%level of significance. The Table 3 reveals the rejections of the null of no cointegration for six of the seven tests at 5 % level of significance Therefore; one may conclude that a long-run money demand exists for the considered panel, as all its variables are cointegrated. 5 I(1) I(1) I(1) 4.2. Long-Run Demand of Money In this section, we proceed to generate individual long-run estimates for equation 2 of demand of money in GCC countries. However, as foreign opportunity costs are roughly identical to domestic opportunity cost, we use the domestic interest only to avoid multicollinearity. We conduct the procedurally “group-mean” panel fully modified OLS estimator (FMOLS) developed by Pedroni (1999b; 2001) since the basic OLS estimator is a biased and unreliable estimator when applied to cointegrated panels. We also employ the dynamic ordinary least squares (DOLS) estimator of Stock and Watson (1993) for aggregated and desagregated panel and the canonical cointegrating regression (CCR) proposed by Park (1992) for the desagregated panel. 4.2.1. Cointegration FMOLS was firstly designed by the work of Philips and Hansen, (1990); and later by Pedroni, (1995); and, Philips and Moon, (1999) to provide optimal estimates of Co-integration regressions. The method modifies least squares to account for serial correlation effects and for the endogeneity in the regressors that result from the existence of a cointegrating relationship (Phillips 1995). FMLOS not only generates consistent estimates of the parameters in relatively small samples, but also controls for potential endogeneity of the regressors and serial correlation. Therefore, FMOLS dominates OLS and ML estimation even in small samples in the presence of cointegration (Phillips, 1992). In order to check the robustness of our results, we therefore re-estimate the GCC money demand function applying dynamic ordinary least squares (DOLS) estimator of Stock and Watson (1993) and the pooled mean group estimation (PMGE) introduced by Pesaran, Shin and Smith (1999). We compare FMOLS estimates with DOLS and PMGE for the group panel and them we conduct canonical cointegrating regression (CCR) estimations for individual countries level. The results of the three-cointegration techniques for the bloc of 6 countries are displayed in table 4. All the three estimators correct the standard pooled OLS for serial correlation and endogeneity of regressors that are normally present in long-run relationship. Table 4. Long Run Money Demand in GCC Countries FMOLS Variable Panel GCC DOLS PMGE LnY Dr Ln E LnY Dr Ln E LnY Dr Ln E 0.537*** 0.06411 -0.041** 0.011636 0.359 0.227683 0.496*** 0.069419 -0.032*** 0.011154 0.229 0.248933 0.616*** 0.001413 -0.004*** 0.003627 0.352 0.010596 *** Denotes the rejection of the null hypothesis at 1%, 5% and 10% level of significance respectively Note. Numbers in italic are the standard deviations Firstly, it appears that FMOLS, DOLS and PMGE outputs reveal consistent and accurate results. As it was expected, income elasticity is positive and significant at 1% level of significance in all three estimations. Moreover, the income coefficient ( 1 0.537, 0.496 and 0.616 ) is approximately in line with the Baumol-Tobin model in which the income elasticity has to be β1 = 0.5 (Baumol, 1952), Tobin (1956). The estimated coefficient of interest rates, which represents the semi-elasticity, is negative and significant at 1% level of significance. This is also in line with standard monetary theory (Friedman 1956) as holding physical assets produce costs. The domestic interest rate represents the opportunity cost of holding money; that interest rates fall when the money supply increases, since the lower interest rates make people more willing to hold the extra cash. However, when interest rates rise, the public would prefer holding more financial assets such as treasury bills, bonds, etc. The recent boom of energy prices was associated with low interest rates have in turn stimulated the GCC economy because affordable money make it more attractive to borrow and to invest and more attractive to spend rather than save. It is worth mentioning that during the previous decade, GCC countries have experienced a buoyant economic growth and the financial sector has witnessed a boom. New financial instruments and policy have been introduced into the GCC financial market and the region has become the hub of finance and insurance industry. Consequently, GCC households tend to adopt more and more new sophisticated interestbearing assets rather than placing their money is saving account. The change in GCC households’ behaviors has simulated money demand in these countries. 6 Regarding exchange rate, it appears to impact positively money demand in GCC countries but not significant. This shows that the appreciation of GCC currencies could raise money demand but it also shows that the substitution effect does not have serious consequences on GCC economies as they have pegged their currencies to US dollar, except for Kuwait. According to the results above, we can conclude that there is evidence of a cointegrating money demand among Gulf Arab countries. This fact is important since there is the project of a GCC monetary Union is under implementation. Turing now to individual country level, the estimation results are based on FMOLS, DOLS and CCR. The results are displayed in the table 5. Table 5. Long Run Money Demand in GCC Countries FMOLS DOLS CCR Variable Bahrain (GDP) LnY 0.615*** 0.126 Dr -0.050* 0.03 Ln E -0.818 0.562 LnY 0.647*** 0.123385 Dr -0.033 0.032331 Ln E -0.158 0.500457 LnY 0.606*** 0.112164 Dr -0.05 0.034405 Ln E -0.844 0.551134 Kuwait 0.968*** 0.102306 -0.008 0.019762 -0.002 0.099245 0.927*** 0.101759 0.002 0.020863 0.021 0.088115 0.832*** 0.085596 -0.013 0.017951 0.08 0.09882 Oman (GDP) 0.577165 0.172664 0.023174 0.02054 1.207942 2.797811 1.263*** 0.198265 -0.043* 0.026708 -0.569 1.602292 1.273*** 0.180169 -0.041* 0.02372 -0.733 1.227877 KSA 0.597* 0.220735 -0.088*** 0.019212 1.199*** 0.130946 0.721** 0.234913 -0.069*** 0.014331 1.267* 0.12454 0.782** 0.216301 -0.088*** 0.019556 1.621* 0.135497 UAE 1.320*** 0.077236 -0.023* 0.015711 1.201 0.13272 1.284*** 0.0859 0.023* 0.017952 1.259 0.150672 1.326*** 0.076778 -0.024* 0.015785 1.169 0.135966 Qatar 0.933*** 0.109416 -0.068* 0.031127 1.462 0.057314 0.934*** 0.052956 -0.054* 0.028601 1.731 0.05453 0.931*** 0.050062 -0.069** 0.027251 1.519 0.04562 *** Denotes the rejection of the null hypothesis at 1%, 5% and 10% level of significance respectively Note. Numbers in italic are the standard deviations The coefficients of FMOLS estimations presented in table 4 show that the income elasticities are positive for all the countries as well as for the panel. These elasticities are ranging from 0.057 for Bahrain to a whopping 1.32 for UAE. The result of an income larger that unit is not surprising. It is even a common finding in both time series and panel data papers on money demand. Income elasticities for Bahrain, Oman and KSA are in line with the Baumol-Tobin model which predict a magnitude of 0.5 for 1 while Kuwait, Qatar and UAE follow the quantitative theory which predict a magnitude of 1 for 1 . Results also show that all the coefficients are significant except for Oman. Similarly, the interest semi-elasticity has the expected sign for the six countries as well for the Panel and its coefficient is ranging from -0.08 for Bahrain to -0.068 for Qatar meaning that there exist an inverse relationship between interest rate and demand for money. The sign is consistent with our postulate. However, it is not significant for Kuwait only. Consistent with the literature on currency substitution, the benchmark money demand function is extended by the real effective exchange rate (E). The few panel data studies including exchange rates produce ambiguous results. The coefficients take values between −1.73 (Rao et al. 2009) and +0.31 (Narayan et al. 2009). In this paper, the results show that the coefficient varies between -0.002 for Kuwait to 1.46 for Qatar. The results of FMOLS and DOLS and CRR are in somewhat identical in our case. All the three procedure reveal on the one hand a positive relationship between output and money demand and a negative relationship between interest rates and money demand on the other hand. Regarding exchange rate, the results diverge among the countries. While it impact negatively money demand in Bahrain and KSA in FMOLS estimations, its coefficient but remains negative for Bahrain but becomes positive for KSA as well as the other countries in DOLS estimation. In CCR, Bahrain and Oman have a negative sign of exchange rate coefficients while the other countries have positive impacts. 4.2.2. Granger Non-Causality Tests: Toda and Yamamoto procedure 7 In this section we will test for causality between the variables of our study by using Granger causality procedure due to Toda and Yamamoto (1995). 4.2.2.1. Granger No-Causality Test Toda and Yamamoto (1995) introduce a method that is used to estimate unrestricted VAR by the use of a Modified Wald test for restrictions on the parameters of the VAR (k) model and estimates a VAR [k+dmax], where k is the lag order of VAR and dmax is the maximal order of integration for the series in the system (Hamdi 2013). The multivariate framework of our case study can be expressed as follows: k d max k d max k d max k d max Md Md 1 1i ln 1i lnYt i 1i Drt i 1i ln Et i 1t P P t i i 1 i 1 i 1 i 1 d k d max k d max k d max k d max M ln NOGt 2 2i ln 2i lnYt i 2i Drt i 2i ln Et i 2t P t i i 1 i 1 i 1 i 1 d k d max k d max k d max k d max M ln Drt 3 3i ln 3i lnYt i 3i Dr 3i ln Et i 3t P t i i 1 i 1 i 1 i 1 (5) (6) ln ln Et 4 Where ln k d max i 1 4i ln Md P t i k d max k d max k d max i 1 i 1 i 1 4i lnYt i 4i Dr 4i (7) (8) ln Et i 4t Md , is the logarithm of real general stock of money, LnY is is the logarithm of real Non-oil P GDP for Kuwait, Qatar, Arabia and UAE and it is replaced by GDP for Oman and Bahrain; Dr is the domestic interest rate and finally E is the exchange rate. The basic procedure of conducting the Toda-Yamamoto method involves two phases. The first step consists in determining the lag length (k) of VAR model and the maximum order of integration (d) of the time series variables in the system. After the selection of optimum lag length VAR (k) and the order of integration dmax, a level VAR is estimated with a total of [k+dmax] lags. The second step requests the application the standard Wald tests on the first (k) VAR coefficient matrix to make Granger causal inference using a chi square (χ2) distribution1. 4.2.2.2. Results We already determined the order of integration of the series (dmax) and we showed that the series are integrated of order one. Now we determine the optimal lag length of the model using the sequential modified LR test statistic (LR), Final prediction error (FPE), Akaike information criterion (AIC), Schwarz information criterion (SC), and Hannan-Quinn information criterion (HQ). The result of selecting optimal lag length of VAR indicates that lag order of VAR (k) is 2, for multitrivariate VAR. Lag 0 1 2 3 4 Log L -2668.939 1220.622 1424.956 1440.378 1456.002 Table 6. Lag Length Criteria for Panel GCC LR FPE AIC NA 1.378113 11.67222 7694.197 6.21e-08 -5.242890 400.6386 2.73e-08* -6.065311* 29.96859 2.74e-08 -6.062787 30.08686* 2.74e-08 -6.061143 SC 11.70827 -5.062677 -5.740929* -5.594235 -5.448420 * indicates lag order selected by the criterion As our series are I(1), this means that, dmax=1. Further, the result of selecting optimal lag length of VAR indicates that lag order of VAR (k) is 2, for multivariate Panel VAR. Therefore, we can estimate a VAR system in levels with a total of dmax+k lags for each country. The results are displayed in table 6. Table 7. Toda and Yamamoto Granger Causality Results 8 LnY => lnM2 Dr => lnM2 E => lnM2 M2 => lnY Dr=> lnY E => lnY LnM2 => Dr LnY=> Dr E => Dr LnM2 => E LnY=>E Dr => E P=K+dmax Panel GCC 9.201*** 4.046** 0.008 4.924** 0.056 182 0.055 0.0765 0.058 0.422 0.0634 0.936 (2) Bahrain 0.184 0.844 0.0105 4.537** 0.433 1.493 3.401** 13.714*** 0.0294 0.353 0.694 0.008 (2) Kuwait 0.702 0.008 1.224 7.576*** 0.129 0.000*** 0.026 0.023 0.312 1.13071 4.470 0.100 (4) Oman 0.0289 0.015 0.0495 2.855** 0.699 1.340 76.48*** 74.371*** 0.536 1.255596 1.010 7.336*** (3) KSA 1.045 2.286 1.681 3.318** 5.175** 0.021 38.147 0.715 1.241 0.652 0.170 1.416 (2) UAE 0.258 0.950 0.011 2.972** 0.131 0.007 0.353 0.026 0.005 0.299 0.014 0.118 (2) Qatar 6.430*** 0.762 1.340 5.016** 0.685 0.415 84.322*** 17.394*** 3.080 1.626 0.038 0.068 (4) *** Denotes the rejection of the null hypothesis at 1%, 5% and 10% level of significance respectively Different interesting conclusions could be drawn from table 7. First, the panel estimation shows the existence of a bidirectional relationship between money demand proxied by M2 and income. This conclusion shows the interdependence between the two variables. A high money demand boosts non-oil GDP sector in GCC which in turn would improve the diversification of GCC economies and would lower their dependency to oil revenue and natural resources rents. During the past few years, GCC governments have undertaken huge structural reforms to diversify their economies as oil and gas are exhaustible resources. The diversification’ strategy differs from country to another one. For example, while Bahrain and Qatar focused on the role of the financial sector and they have become hubs of finance in the region, UAE have focused their diversification in infrastructure to attract FDI and tourism. Oman also has become the preferred destination for luxury tourism while KSA is the preferred destination of large international manufactories. Second, the result shows also a unidirectional relationship running from domestic interest rate to money demand M2. This is in line with the Quantity Theory of Money (QTM), which indicates the negative relationship between money demand and interest rate. This result also confirms our finding in table 4. 5. Conclusion and Recommendations In this paper, we estimated and analyzed the aggregate and individual long-run money demand functions of the six Gulf Cooperation Council countries during the period 1980Q1-2011Q4. After checking for stationarity of the different variables of the model using Panel unit root test, we applied the panel FMOLS, PDOLS and PMGE method to estimate the long-run money demand function for six Gulf Cooperation Council countries. From an aggregated analysis, we found that income elasticity is around 0.5 ( FMOLS 1 0.537, DOLS1 0.496 and PMGE 1 0.616 ) which is in line with the Baumol-Tobin model in which the income elasticity has to be β1 = 0.5 (Baumol, 1952), Tobin (1956). The estimated coefficient of interest rates, which represents the semi-elasticity, is negative and significant at 1% level of significance ( FMOLS 2 0.04, DOLS 2 0.03 and PMGE 2 0.04 ). This is also in line with standard monetary theory (Friedman 1956) as holding physical assets produce costs. The empirical literature using aggregated time series data. Panel cointegration tests provided evidence in favor of a stable long-run money demand function. Moreover, similar results were found in the disaggregated analysis (individual countries). The Granger non-causality test due to Toda and Yamamoto (1995) procedure shows evidence of a bidirectional causal relationship between money demand and income. At an individual level, the unique common results between the countries are the evidence of a unidirectional causality running from M2 to income. Finally, the overall results show that exchange rate does not affect long-run money demand functions of the six Gulf Cooperation Council countries. The purpose of this study is to demonstrate the importance of money demand in conducting a sound monetary policy because 9 the central banker in GCC countries would needs to make sure the elasticities are stable throughout time. This is one of the several requirements of a successful monetary Union. The stability of the money demand function plays a central role for the importance of money for the monetary policy; especially because the GCC countries are moving toward a single currency managed by a single central Bank. The goal of having union monetary policy strategies in GCC countries would support the price stability because many of those countries have faced an increase in inflation since 2002, which was accompanied with oil price boom. In fact, inflation decreases the purchasing power of consumers in the GCC countries. This research is important in this period because the GCC countries are trying to move toward creating of a Monetary Council and a single currency. References Arango, S., and M. I. Nadiri (1981) Demand for Money in Open Economies. Journal of Monetary Economics 7: 69-83. Baba Y, Hendry DF, Starr RM. 1992. The demand for m1 in the USA, 1960-1988. The Review of Economic Studies 59: 25–61. Baharumshah, A.Z., Mohd, S.H., Yol, M.A. (2009). Stock prices and demand for money in China: New evidence, Journal of International Financial Markets, Institutions and Money, 19(1), 171-187. Bahmani-Oskooee, M. (1996) The Black Market Exchange Rate and Demand for Money in Iran. Journal of Macroeconomics 18: 171-176. Bahmani-Oskooee, M. and A. Techaratanachai, (2001), “Currency substitution in Thailand,” Journal of Policy Modelling, 23: 141-145. Bahmani-Oskooee, M., Rehman, H., (2005). Stability of the Money Demand Function in Asian Developing Countries, Applied Economics, 37, 773-792. Bahmani-Oskooee, M., Gelan, A., (2009). How stable is the Demand for Money in African Countries, Journal of Economic Studies, 36(3), 216-235. Ball L. (2001). Another look at long-run money demand. Journal of Monetary Economics 47: 31–44. Basher, S., Fachin, S., (2012). Investigating Long-Run Demand for Broad Money in the Gulf Arab Countries, DSS Empirical Economics and Econometrics, Working Papers Series, DSS-E3 WP 2012/6, La Spienza University. Baumol WJ. 1952. The transactions demand for cash: An inventory theoretic approach. Quarterly Journal of Economics 66: 545–556. Boone L, van den Noord P. (2008). Wealth effects on money demand in the euro area. Empirical Economics 34(3), 525–536. Breitung, J. (2000) “The Local Power of Some Unit Root Tests for Panel Data”, in Baltagi, B. (ed.), Nonstationary Panels, Panel Cointegration, and Dynamic Panels, Advances in Econometrics, Vol. 15, JAI: Amsterdam, 161-178. Cheong, C., (2003). Regime changes and econometric modeling of the demand for money in Korea, Economic Modelling, 20(3), 437–453. Darrat, A. and Mutawa, A. (1996) Modeling money demand in the United Arab Emirates, The Quarterly Review of Economics and Finance, 36(1), 65–87. Darrat, A.F. and Al-Sowaidi, S.S. (2009). Financial progress and the stability of long-run money demand: Implications for the conduct of monetary policy in emerging economies. Review of Financial Economics 18, 124–131. De Bondt, GJ., (2009). Euro area money demand: Empirical evidence on the role of equity and labour markets. ECB Working Paper No. 1086. Dickey D.A., D.Jansen,D. Thronton "A Primer Cointegration with an Application to Money and Income" Federal Reserve Bank of St. Louis, (March-April 1991) :58-77. Dreger C., Wolters J. (2010). Investigating M3 money demand in the euro area. Journal International Money Finance 29(1),111–122. Duca JV, VanHoose DD. (2004). Recent developments in understanding the demand for money. Journal of Economics and Business 56, 247–272. Engle, R.F., Granger, C.W.J. (1987), “Cointegration and error correction: representation, estimation, and testing”, Econometrica, 55, 251–76. 10 Friedman, M. (1956).' The Quantity Theory of Money: A Restatement" In Studies in the Quantity Theory of Money, edited by M. Friedman. Chicago: University of Chicago Press. Friedman, M. (1959). The demand for money: some theoretical and empirical results, Journal of Political Economy, 327-351. Friedman, M., Schwartz, A.J. (1982), Monetary trends in the United States and the United Kingdom, Chicago, University of Chicago Press. Foresti and Napolitano (2012). Modeling Long-run Money Demand for Nine Developed Economies. CRISEI Discussion Paper No. 20 Novembre 2012 ISSN: 2280-9767 Hamdi H, (2013), Testing Export-led Growth in Tunisia and Morocco: New Evidence using the Toda and Yamamoto procedure “Economics Bulletin, 2013, Vol. 33 No. 1 pp. 677-686 Hamdi H and Sbia R. (2014), The dynamic relationship between CO2 emissions, energy usage and Growth in Gulf Cooperation Council (GCC) countries: An aggregated analysis. Economie Appliquée. Vol. 67.2014, 2, p. 161-182 Hamdi, H Sbia, R. and Onur Tas B (2014). Financial deepening and Economic Growth in GCC. International Economic Journal. Vol. 28, Iss. 3, 2014. Routledge Taylor & Francis Online Harb, N., (2004). Money demand function: A heterogenous panel application. Applied Economics Letters, 11, 551–555. Hayo, B., (1999). Estimating a European Demand for Money. Scottish Journal of Political Economy, 46(3), 221-44. Hoffman DL, Rasche RH. (1991). Long-run income and interest elasticities of money demand in the United States. The Review of Economics and Statistics 73(4): 665–674. Im, K. S., M. H. Pesaran, and Y. Shin (2003). Testing for unit roots in heterogeneous panels. Journal of Econometrics 115 (1), 53-74. International Monetary Fund. (2013). World Economic Outlook, April 2013: Hopes, Realities, Risks (World Economic & Financial Surveys). Retrieved from http://www.imf.org/external/pubs/ft/weo/2013/01/pdf/text.pdf. Jawadi, F., M. Sousa, R., (2013). Money demand in the euro area, the US and the UK: Assessing the role of nonlinearity, Economic Modelling, 32, 507–515. Johansen, S., (1988). Statistical analysis of cointegration vectors, Journal of Economic Dynamics and Control, 12, 231-54. Kaldor, N. (1982), The Scourge of Monetarism, London, Oxford University Press. Kao, C. (1999). Spurious regression and residual-based tests for cointegration in panel data. Journal of Econometrics 02/; 90(1):1-44. Kao, C. and Chiang, M.-H. (2000). On the estimation and inference of a cointegrated regression in panel data, in Baltagi, B. (Ed.) Nonstationary Panels, Panel Cointegration, and Dynamic Panels, Advances in Econometrics, Vol. 15, JAI Press, Amsterdam, pp. 161–178. Khan, M.S. (2009). The GCC Monetary Union: Choice of Exchange Rate Regime. Working Paper no. 09-1. Peterson Institute for International Economics, Washington, DC. Khatib-Kswani, M., Towaijari, H. A., (1999). Cointegration Error Correction and the Demand for money in Saudi Arabia, Economia-Internazionale, 52(3), 299-308. Laidler, D.E.W. (1977). The Demand for Money: Theories and Evidence, second edition, London, Harper & Row. Laidler, D.E.W. (1982). Monetarist Perspectives, London, Philip Allan. Laidler DEW. (1985). The Demand for Money: Theories, Evidences, and Problems. Harper & Row, Publishers, New York. Lee, C.C., Chien, M.S., (2008). Stability of money demand function revisited in China, Applied Economics, 40(24), 3185-3197. Lee, C-C., Chang, C-P., Chen, P-F., (2008). Money demand function versus monetary integration: Revisiting panel cointegration among GCC countries. Mathematics and Computers in Simulation 79, 85–93. Levin A, Lin C, Chu CJ (2002). Unit root tests in panel data: asymptotic and finite-sample properties. JEconom 108(1):1–24 Maddala GS, Wu S (1999). A comparative study of unit root tests with panel data and a new simple test. Oxf Bull Econ Stat 61:631–652. 11 Mark N C. and Sul D (2003). Cointegration Vector Estimation By Panel Dols And Long-Run Money Demand, Oxford Bulletin Of Economics And Statistics, 65, 5 (2003) 0305-9049. Mehra YP. (1993). The stability of the m2 demand function: Evidence from an error-correction model. Journal of Money, Credit, and Banking 25(3): 455–460. Miller SM. (1991). Monetary dynamics: An application of cointegration and error-correction modeling. Journal of Money, Credit, and Banking 23(2): 139–154), Mundell R.A (1963). Capital Mobility and Stabilization Policy under Fixed and Flexible Exchange Rates”, Canadian Journal of Economics and Political Science, 29, Nov. 1963, pp. 475-485. Park, J., (1992). Canonical cointegrating regressions, Econometrica 60, 119-143. Pedroni P., (1999). Critical values for cointegration tests in heterogeneous panels with multiple regressors. Oxford Bulletin of Economics and Statistics Special Issue 653–670. Pedroni P., (2004). Panel Cointegration, Asymptotic And Finite Sample Properties Of Pooled Time Series Tests With An Application To The PPP Hypothesis. Econom Theory 20:597–625 Pedroni P., (2000) Fully Modified OLS For Heterogeneous Cointegrated Panels. In: Advances in econometrics, vol 15. Emerald (MCB UP), Bingley, pp 93–130. Phillips PCB, Perron P (1988) Testing for a unit root in time series regression. Biometrika 75(2):335– 346. Phillips PCB, Hansen BE. (1990). Statistical inference in instrumental variables regression with i(1) processes. Review of Economics Studies 57: 99–125. Phillips, P. C. B. and Moon, H. R. (1999). ‘Linear regression limit theory for nonstationary panel data’, Econometrica, Vol. 67, pp. 1057–1111. Rao, B.B. and Kumar, S. (2007) ‘Structural breaks, demand for money and monetary policy in Fiji’, Pacific Economic Bulletin, 22, 53-62. Rao, B. B., A. Tamazian, and P. Singh (2009). Demand for money in the Asian countries: A systems GMM panel data approach and structural breaks. MPRA Paper No. 15030. Stock JH, Watson MW. (1993). A simple estimator of cointegrating vectors in higher order integrated systems. Econometrica 61(4): 783–820. Tobin J., (1956). The interest elasticity of transactions demand for cash. Review of Economics and Statistics 38: 241–247. Toda, H. Y., and T. Yamamoto (1995). Statistical Inference in Vector Autoregressions with Possibly Integrated Processes. Journal of Econometrics 66, 225–50. Valadkhani A, Alauddin M., (2003). Demand forM2 in developing countries: an empirical panel investigation. School of Economics and Finance Discussion Paper No. 149, Queensland University of Technology. World Bank (2012). World Development Indicators: CD-ROM 2004. Washington, D. C.: The World Bank. Wu, Ge, (2009). Broad Money Demand and Asset Substitution in China, IMF Working Papers, 09/131 (Washington: International Monetary Fund). 12

© Copyright 2018